Using a DSO For Signal Analysis

The oscilloscope has been a primary tool for electronic design engineers since it was invented many years ago. The first decades of oscilloscopes were analog in nature. Their fundamental technology was the front-end amplifier, sweep generator, and—most particularly—the phosphor used to coat the screen of a CRT. That phosphor served as a memory element that briefly held the shape of the signal on the CRT for the viewer.

The value of the scope was its capability to trigger many times per second and overlay the phosphorescent images on the screen. Information concerning the wave shape of the signal was transferred via viewing the signal (sometimes using a hood to eliminate light sources and at other times using a camera). The analysis was done in the brain of the human who viewed the information and extracted insights from the wave shape.

Back in the early 1980s, the analog oscilloscope began to give way to a new type of instrument for capturing and measuring signals—the digital oscilloscope. The digital scope, sometimes called the digital storage oscilloscope (DSO), offered the user the ability to capture a waveform by converting the analog data to digital numbers, then displaying the data points on the CRT screen.

Because all the data was stored in memory, there were several advantages to this technology with regard to viewing and analyzing a waveform. Engineers looking at single-shot or low-repetition-rate events found that the DSO provided a way to capture, store, and view a very brightly displayed waveform regardless of the rep rate. Also, the DSO did not suffer from the waveform decay or blooming display issues that were problematic with analog storage oscilloscopes.

The DSO’s stored data gave users the ability to dive further into analysis of the waveform than earlier viewing technologies. New tools to perform waveform measurements automatically increased measurement accuracy and removed subjectivity and variability caused by the engineer reading the screen.

There were feature sets like pre- and post-trigger display, multiple zoom capability, and waveform analysis such as averaging, basic math, and FFT that the basic analog scope could not do. These tools made waveform analysis available to every engineer without the need to use stand-alone digitizer cards and write software to analyze the data.

Some applications still are best suited to oscilloscopes that can quickly trigger and overlay many signals on the screen simultaneously. The fastest analog storage scope on the market today can trigger up to 1,000,000 times/s and draw each signal on the screen in real time, with a variable decay rate for the phosphorescence.

But the market for such scopes is dying, primarily because design and test engineers no longer can extract sufficient information from a signal by viewing it. In many applications, viewing tools are becoming a dead end. They take the user a short distance in the right direction, but can’t give the information really needed.

As an example, an engineer might be tasked with verifying the performance of a 133-MHz clock used to transfer data to and from a microprocessor. One key question is whether the clock meets certain test standards for cycle-to-cycle jitter. Imagine an engineer trying to observe 133 million cycles of the waveform per second on the screen and deciding by eyeball if the clock meets specification. Impossible.

This brings us to the concept of wave-shape analysis, the capability of an oscilloscope to re-present complex data using a format which is not the usual voltage vs. time waveform. The new view of the data should allow the engineer to discern by viewing and confirm by measuring using the raw signal shape and putting cursors on it to make a measurement.

This type of analysis is based on being able to view waveforms in the time, frequency, statistical, and parameter modulation domains. Only by using this broad range of views can the oscilloscope user gather a complete understanding of the circuit/device performance.

Wave-shape analysis involves the capability to use powerful math functions to transform signals from one domain to the other. All of this processing has to be easily accessible, understandable, and lightning fast, otherwise it will not be used.

The classic oscilloscope measurement provides a view of voltage as a function of time. While this basic measurement tool has stood the test of time, today’s complex signals require much greater wave-shape analysis capabilities than simple cursor or parameter measurements to extract the useful information. It has become difficult to confirm signal behavior only by using conventional parameters because many measurements of this type cannot effectively or completely operate on long, complex data sets.

Figure 1 shows a simple example of wave-shape analysis. Often, the key characteristics of a signal are computed as pulse parameters such as frequency, duty cycle, or the timing skew between two edges.

Early digital scopes simply would display the latest measurement of the parameters. Several years ago, oscilloscopes began displaying statistics including the high, low, average, and rms values of parameters. But the user gets little insight into the source of circuit faults from these numbers. The scopes simply report if the signal meets or fails the specification.

A histogram is a statistical view of the parameter data. It is a bar chart that shows how often each value of the parameter occurred. This new view, a re-presentation of the data, allows simple viewing and easy measurements that extract information from a complex set of raw data. The histicon, an icon-size view of a histogram of the frequency parameter in Figure 1, shows the basic frequency is not stable, and the bathtub shape is an immediate indication that the frequency variation is due to a sinusoidal modulation.

The Gaussian shape of the duty-cycle histogram means this parameter has a central value affected by noise. The skew between channel 3 and channel 4 is a flat histogram. This indicates there is an equal chance of timing skew, over a certain range of times, between the two channels. The user gets immediate information from a straightforward view of each distribution and can make simple measurements. The most advanced oscilloscopes have very fast data throughput that enables the display of up to eight simultaneous histicons of any parameter.

The capability to extract useful information from complex signals is further complicated by the length of the waveform record. The challenge is to create a DSO with high sampling rates, long memory, and a very specialized hardware/software infrastructure that has been designed to acquire, move, process, and extract useful information from long, complex data records. This process must be fast so the engineer is not kept waiting for the process to complete and the usability of the instrument stays high.

A new, fast, streaming architecture named X-Stream technology makes measurements 10 to 100 times faster than previously possible. It also provides the capability to decrease dead time when making measurements, increasing the likelihood of measuring an intermittent fault.

Example: Characterizing Performance of a PLL

Let’s look at an example of a moderately complex measurement—the characterization of performance of a phase-locked loop (PLL). In Figure 1, a statistical domain view of signal characteristics gives insight into the waveform.

A different type of wave-shape analysis in the time domain is shown in Figure 2. The acquired waveform is 200 µs long and sampled at 8 GS/s, generating 1.6 million points of data. The upper trace is 100,000 cycles of a PLL. The lower trace is formed from a series of numbers that tracks the frequency of the PLL.

During the acquisition window, the PLL is kicked from a low frequency to a higher one. While the upper trace presents the usual voltage vs. time oscilloscope display, the lower trace is a new presentation of the data. Every period of the PLL is measured, and the lower trace tracks the PLL frequency (one value per period for each of the 100,000 periods) vs. time.

The new view of the data shows that the frequency begins at a stable, low value, then there is a step that overshoots and settles to a new, higher frequency. We call this trace a JitterTrack. It is easy to view the timing changes of a circuit with this type of trace and make measurements such as the amplitude of the frequency shift (70.2 kHz), base/top frequencies (9.638/10.0340 MHz), rise time of the frequency step (2.50112 µs), and overshoot (62.32 %).

The PLL response shown in Figure 2 is fairly simple. In general, the output phase of a PLL will respond to changes in the input phase, but only if those changes are within the bandwidth range of the PLL. Input changes that occur at low frequencies are passed to the output, but high-frequency changes are too fast for the PLL to respond.

An engineer often will want to characterize the response curve of a PLL vs. frequency. This sometimes is called the PLL loop bandwidth or the jitter transfer function of the PLL.

This brings us to a third type of view that can be used to gain insight into circuit behavior: the view in the spectral/frequency domain. We can measure the PLL loop bandwidth by applying an input signal that contains a step change in phase. This will allow us to measure the step response of the PLL. The PLL impulse response can be obtained by differentiating the step response. The FFT of the impulse response is the frequency response of the PLL.

Figure 3 shows the measurement of the frequency response of the input signal. The top trace (left side) is the input reference signal, a 66.67-MHz signal with a 2-radian step in the phase at the center of the trace.

The JitterTrack (second trace on left) of time interval error (TIE) actually shows the step. This wave-shape analysis tool, Parameter Modulation View, takes the large and complex data set with a phase shift in the top trace and re-presents it in a simple-to-discern and measure processed trace.

We have just extracted very useful information from the input signal. This is differentiated (third trace left) and then processed through an FFT and averaged to show the frequency response of the input in the bottom trace. This is the input excitation to the PLL, and it is a spectrally flat input out to 5 MHz to within about ± 1.5 dB.

The output signal is analyzed as shown in the right side of Figure 3. Going through the identical processing steps, the result is the output response of the PLL. It is a low-pass characteristic with an upper cutoff frequency of about 3 MHz. There is a broad peak at about 2.5 MHz, after which the response rolls off.

The final step is to divide the output spectrum by the input spectrum as shown in Figure 4. As these are both logarithmically weighted, this is the normalized frequency response of the PLL.

All of the wave-shape analysis we have applied has been derived from the raw voltage vs. time measurements of the input and output waveforms. But, as in many applications, the user cares less about the raw data and is more concerned with information concerning component/circuit performance.

These crucial performance characteristics can best be observed from waveform analysis in the time, frequency, statistical, and parameter modulation domains. Current oscilloscopes, which include these highly integrated wave-shape analysis tools, more than pay for themselves when it comes to solving problems involving today’s complex electronic circuits and devices.

About the Authors

Dr. Michael Lauterbach is director-product management at LeCroy. He has worked more than 20 years for LeCroy, starting as manager of engineering services. His doctorate from Yale University is in high-energy physics. Dr. Lauterbach has published more than 30 papers on the use of digital test equipment and presented seminars at technical conferences and for engineers at IBM, Motorola, Seagate, and the U.S. State Department. e-mail: [email protected]

Wayne Swirnow is the sales engineer in the Metro New York territory for LeCroy. During his 23-year career with the company, he has held a variety of positions in engineering, service, sales, marketing, and operations. e-mail: [email protected]

LeCroy, 700 Chestnut Ridge Rd., Chestnut Ridge, NY 10977, 845-425-2000.

Return to EE Home Page

Published by EE-Evaluation Engineering
All contents © 2002 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.

July 2002

Sponsored Recommendations

Near- and Far-Field Measurements

April 16, 2024
In this comprehensive application note, we delve into the methods of measuring the transmission (or reception) pattern, a key determinant of antenna gain, using a vector network...

DigiKey Factory Tomorrow Season 3: Sustainable Manufacturing

April 16, 2024
Industry 4.0 is helping manufacturers develop and integrate technologies such as AI, edge computing and connectivity for the factories of tomorrow. Learn more at DigiKey today...

Connectivity – The Backbone of Sustainable Automation

April 16, 2024
Advanced interfaces for signals, data, and electrical power are essential. They help save resources and costs when networking production equipment.

Empowered by Cutting-Edge Automation Technology: The Sustainable Journey

April 16, 2024
Advanced automation is key to efficient production and is a powerful tool for optimizing infrastructure and processes in terms of sustainability.

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!